HAL Id: inserm-00086746
https://www.hal.inserm.fr/inserm-00086746
Submitted on 12 Apr 2007HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Socioeconomic inequalities in premature mortality in
France: have they widened in recent decades?
Annette Leclerc, Jean-François Chastang, Gwenn Menvielle, Danièle Luce
To cite this version:
Annette Leclerc, Jean-François Chastang, Gwenn Menvielle, Danièle Luce. Socioeconomic inequalities in premature mortality in France: have they widened in recent decades?. Social Science and Medicine, Elsevier, 2006, 62, pp.2035-45. �10.1016/j.socscimed.2005.08.031�. �inserm-00086746�
Socioeconomic inequalities in premature mortality in France: have they widened in recent decades?
LECLERC A1, CHASTANG1 JF, MENVIELLE G, LUCE D and the EDISC Group
1 INSERM U687, IFR69, Saint-Maurice, France
Abstract
An increase in social inequalities in premature mortality has been observed in the last decades
in various European countries. In France large inequalities have been reported for several
years, but the changes over time have been only partially examined.
The study was based on the analysis of a large longitudinal data set. Four periods of seven
years following a census were compared. Deaths in the period (21003 deaths for men, 9418
for women) were recorded and studied according to the socioeconomic status (SES) at the
census. Relative index of Inequality (RII) were calculated in order to quantify the magnitude
of inequalities among those employed, and also in the entire population, with specific
categories for those inactive. The Results showed that the magnitude of inequalities remained
rather stable over time for men and women working at the time of the census. However, for
the entire population, a strong increase in the magnitude of social inequalities was observed.
For men the RIIs increased from 3.53 in the first period to 6.54 in the most recent period. For
women the corresponding figures were 1.94 and 3.88. The increase was observed also for
specific causes of deaths: cancer and cardiovascular diseases for both sexes, and external
causes for men. In spite of a global decrease of the mortality over the period, the absolute
differences between the top and the bottom of the socioeconomic scale did not change.
The results highlight the importance of temporal changes in mortality associated with an
increase of unemployment, changes in the labour market, and consequences in selective
Keywords : mortality, inequality, social, employment.
Word count for the abstract: 279 words Full word count : 5433 words
Several reports document an increase in social inequalities in premature mortality in recent
years in various European countries (Mackenbach & al.,2003; Donkin, Goldblatt & Lynch,
2002; Marang-Van de Mheen, Davey Smith, Hart & Gunnings-Schepers, 1998; Davey Smith,
Dorling, Mitchell & Shaw, 2002). In France, large inequalities have been reported for several
decades (Mackenbach et al.,1997; Kunst, 1997; Lang, Fassin, Grandjean, Kaminski &
Leclerc, 2002; Leclerc, Lert & Fabien, 1990). Standardized mortality ratios (SMRs) for men
according to socioeconomic status (SES) have been published for several periods (Mesrine,
1999), and other analyses have focused on those looking for work (Mesrine, 2000). The
changes over time have been described for various socioeconomic groups (Monteil &
Robert-Bobée, 2005), and for the elderly (Cambois, 2004). However, these analyses describe only
partly changes over time, especially because the size of socioeconomic groups, that is, the
distribution of SES, in France has changed substantially since 1968. Unemployment had also
increased.
Several studies stress the role of occupational status for mortality; in addition to social
position based on present or past occupation, being employed or not may have effects on
mortality (Mesrine, 2000). Relations between employment and health can be viewed also in
the other direction: in the last decades the role of health selection process towards inactivity
may have increased in several countries (Virtanen & coll, 2003; Lissau & coll, 2001;
Cardano, Costa & Demaria, 2004; Sermet & Khlat, 2004; Bartley & Ferrie, 2001).
The objective of this study was to describe mortality over time in France, with a specific
interest for those inactive, using methods that take into account mortality in this group, and
also the size of the socioeconomic groups (Regidor, 2004). We compared the magnitude of
socioeconomic inequalities, separately for men and women aged 30-64 years, for four
Methods
Permanent Demographic Sample
These data come from a longitudinal national dataset kept by INSEE (the French census and
statistics bureau). It is known as the EDP (Echantillon Démographique Permanent, or
permanent demographic sample) and it currently contains approximately 1 million persons
(roughly 1% of the population) randomly selected according to their date of birth (4 days
during the year). Inclusion occurs by birth, by marriage (for foreigners with the appropriate
birthday if they marry a French national and live in France with them), or at a census,
especially for immigrants (Rouault, 1994; Heran, 1998). Once included, a person remains part
of the sample until death. INSEE follows EDP subjects over time, collecting for each of them
social, demographic, and occupational information from different census schedules and from
vital and civil status forms. This information allows an assessment of their SES at the time of
censuses. This dataset is similar to the Longitudinal Study in the United Kingdom, and to
other longitudinal studies in Europe; a list of the main publications relating to the EDP can be
found on the website of the Longitudinal Study
(http://www.cls.ioe.ac.uk/Ls/lshomepage.htlm).
Cause-specific mortality
Information about causes of deaths came from the INSERM department in charge of coding
the causes of deaths. Causes of deaths could be retrieved for 98% of deaths. Causes were
classified according to the WHO guidelines, ICD-8 or ICD-9 depending on the period. Four
large groups of causes of death were taken into account in the study: cancer, cardiovascular,
external causes, and other causes. However, due to small number of deaths in some categories
for women, death by external causes was studied only for men.
SES at the time of the census was recorded according to the French occupation and socio-professional categories (PCS, “profession et catégorie socio-professionnelle”), the standard classification for assessment of SES in France (Chenu, 2000). Seven categories were used:
professionals and managers, farmers, the self-employed (except farmers), intermediate
white-collar workers, office and sales personnel, skilled manual workers, unskilled workers.
Those not working at the time of the census were classified in one of the three following
categories: retired or early retired, job-seekers, and other inactive (including housewives). In
France, those unable to work due to health reasons are in the category “other inactive”. They are not classified as “retired” since health retirement is not an administrative category.
Measure of inequality, relative Index of Inequality
The initial analyses for each period were based on Standardised Mortality Ratios (SMRs) and
Cox models with age as the time variable in order to control for age, and SES as a qualitative
variable.
Relative index of inequality (RIIs) were calculated in order to obtain quantitative global
estimates of the magnitude of inequalities in mortality.
This index is a regression based summary measure widely used in research into social
inequalities (Marang-van de Mheen & al., 1999; Davey-Smith & al., 2002; Mackenback &
kunst, 1997; Singh-Manoux, 2005; Khang, 2004). It is calculated by ranking socioeconomic
categories on a scale from the highest (0) to the lowest (1). Each category covers a range on
the scale proportional to its population size and is given a value on the scale corresponding to
the midpoint of its range (an example is given in figure 1). These data are fitted by a
regression model (here, a Cox model) that yields mortality estimates for the entire social
distribution. The estimates for the two extremes are used to calculate the RII, which is
With figure 1 as an example, the RII gives an estimate of the ratio between mortality at the abscissa “1” and mortality at the abscissa “0”. The value given by the model may be slightly different from the ratio between the two extreme categories, because it takes into account the
size of the groups and the whole distribution of deaths.
Calculating RIIs allowed to take into account the number of subjects in each category, and the
changes in the distribution of SES between periods. The magnitude of inequalities could be
compared between periods, and the hypothesis of stability over time could be tested.
Socioeconomic categories were ranked according to mortality rates in the models, although
that meant different rankings for different periods.
Comparisons over time
The main part of the analysis was the comparison of four sub-cohorts. The baseline for each
was a different census, 1968, 1975, 1982 and 1990, with a follow-up of seven years. Each
sub-cohort comprised men and women in the EDP aged 30-64 years at the relevant census.
Deaths in the follow-up period were recorded and studied according to SES at the census.
Foreigners and those not born in France were excluded because of the incompleteness of the
information on their vital status, especially for foreigners who died abroad. A total of 30421
deaths were included in the study, 21003 for men and 9418 for women. Table 1 reports the
number of subjects and deaths in the cohorts. Table 2 describes changes over the period in the
percentage of active and inactive, with details for those not in the labour force.
The initial analyses suggested that the relation between relative socioeconomic position and
mortality was not linear but closer to exponential, especially if job-seekers and those not in
the labour force were included. For that reason, the next analyses used Cox rather than linear
Mortality in each 7-year period was calculated for each sub-cohort and each sex according to
the SES defined at the census. The RII was calculated for each period. Three sets of analyses
were performed for total mortality:
Comparison 1 – Analyses restricted to those employed at the census, SES in seven categories.
Comparison 2 – Analyses of the entire sample, based on work status at the census, in nine
categories: seven SES categories for those working, plus two additional categories: retired;
job-seeking or out of the labour force (combined for men because in the first periods very few
men were in these two categories).
Comparison 3 – Analysis of the entire sample, based on all available information: previous
SES for those retired and seeking work, present SES for those currently working. Eight categories were compared: seven SES, plus “other unoccupied” for those who could not be classified into a SES category.
For cause-specific mortality, comparisons were based on the situation at the census, described above as “comparison 2”.
To test the hypothesis that socioeconomic inequalities had increased, and to take into account
the decrease in mortality over time, we analyzed total mortality for the entire 1968-1996
period with two Cox models, one for each sex. The mortality risk of a subject was estimated
by the model, taking into account age, period (four period variables), and work status at the
census as a quantitative variable (relative situation, from 0 to 1, as in the previous models for “comparison 2”). The model gave an estimate of the mortality for an “average” subject in each period, and an estimate of the effect of the socioeconomic situation. The inequality
estimate was a single RII, corresponding to the hypothesis that the RII did not change
between periods.
In order to test the alternative hypothesis that the RII had increased, this model was compared
In addition, slightly different models were used to estimate the change in the magnitude of
the RII corresponding to an increase of one calendar year.
These models enabled us to estimate not only ratios or RIIs, which are relative measures, but
also differences. The magnitude of mortality differences between the periods was calculated
(Mackenback & Kunst, 1997; Bostrom & Rosen, 2003). Details on the method are given in
the appendix.
Results
Mortality according to SES
For men the SES rankings according to mortality were stable over time. Of those working at
the census, the scale ranged from professionals and managers to unskilled workers. The main
change was among those classified as retired, job-seeking, or otherwise not in the labour
force, as illustrated in figure 1 for the two extreme periods. Around 1970 these categories
accounted for approximately 11% of all men (table 2); their mortality rate was high: with the
professionals and managers as the reference, their standardised mortality ratios were between
300 and 400. In 1990 22% of the male population belonged to these categories, and the
mortality in these subgroups had increased in comparison with professionals and managers.
The mortality for unskilled workers compared to professionals and managers did not change
much between periods. However, the size of the groups changed over time: in 1968 the male
population was composed of 7% professionals and managers and 17% unskilled workers; in
1990 these percentages were 13% and 7%, respectively.
For women (figure 2) SES rankings were less stable; these rankings were based, however, on
relatively small numbers in some groups. For example, in the first period there were only 22
deaths among 1339 skilled workers. Among job-seekers, the numbers of deaths were smaller
those not employed at the time of the census (job-seeking, retired, or other inactive), when
age was controlled for. The mortality ratios in these categories, with professional and
managers as the reference, were smaller than for men. Over this thirty-year period, the
percentage of women out of the labour force decreased dramatically ( table 2 and figure 2).
The mortality in this group was higher than that of the comparable general population for all
four periods, but the gap between them and the professionals and managers increased with
time, since the SMR rose from 180 to 230.
Magnitude of inequalities in the four periods, total mortality
As Table 3 shows, the magnitude of inequalities, as quantified by the RII, remained rather
stable over time among men employed at the time of the census. However, results based on
the situation at the census (employment status) for the entire male population indicated an
increase of inequalities from 1968 through 1996. Using the third comparison option (entire
sample, last or present SES), the RII was 2.79 for the first period and 2.79, 2.47 and 3.85
thereafter.
The magnitude of inequalities among women was smaller. As for men, comparisons restricted
to the employed were stable over the three decades, but inequality increased when the entire
population was considered and classified according to the situation at the census. The third
option for comparisons yielded an RII of 1.66 for the first period and 1.58, 3.67 and 2.99 for
those that followed.
Inequalities by cause-specific mortality
The RIIs according to the situation at the census (“comparison 2”) were also calculated for
specific causes of deaths: cancer and cardiovascular diseases for both sexes, external causes
(only for men), and for other causes. Results are shown in table 4. In both sexes, the largest
indicated that inequalities had increased over time for the main causes of death. Inequalities were very large for “other causes of deaths”, with an increase over time except between the third and the fourth period for women.
Global changes from 1968 through 1996
Table 5 presents the changes over the entire period, according to the Cox models, for the
entire population, classified by their situation at the census (“comparison 2”). Estimates given
for the four periods are for a person with the average SES; they indicate that age-standardized
mortality decreased during this period, especially for women. In this model the RII is shown
for the first period; for the other periods, this figure must be multiplied by the corresponding
interaction term. All the interaction terms (except 1975 for women) were significant, which
means that socioeconomic inequalities (as defined and measured in the model) increased
significantly. A time gradient was observed for men; for women, however, the increase took
place primarily between the second and the third period. A slightly different model estimated
that the RII was multiplied by 1.017 each year (95% CI 1.010 to 1.023) for men, and by 1.018
(95% CI 1.008 to 1.027) for women.
Figure 3 illustrates the changes in the period according to the models in Table 5. The
reference (100) corresponds to an average SES in the 1968-1974 period; the mortality rate for
an average SES decreased over time, as it did for those at the top of the socioeconomic scale.
At the bottom, however, mortality for men in the last period was not much better than during
the first and remained much worse than that of those at the top of the social scale twenty years
earlier.
The magnitude of inequalities in terms of differences were calculated with these models (see
details on the method in the appendix). For men in the first period the estimate for mortality at
the top of the social scale was 50.4, with as reference (100) mortality for the average SES; it
between the top and the bottom of the social scale was 148. The absolute difference for the
most recent period was 147.5. For women, the absolute differences were 80.4 for the first
period and 77.5 for the most recent period. These comparisons indicate that although
mortality decreased over the period, the level of absolute inequalities remained quite stable.
Discussion
This study was able to compare several periods rather precisely, although the paucity of
women's deaths in some categories is a limitation. An advantage of census data is the quality
of the data and especially the quality of the coding for SES. The exhaustiveness of deaths in
this sample was checked by comparing the mortality rates to the published mortality data and
could be considered good. The main reason for comparing to the published mortality rates
was that we had excluded foreigners and those not born in France. The impact of this decision
is probably limited, which is consistent with previous results on mortality of immigrants in
France (Brahimi, 1980).
The method, mainly based on RII comparisons, overcomes the difficulties due to changes in
the distribution of SES categories over time. It could also provide global measures with
different options for classifying those not working at the census. The method is similar to that
used in a relatively recent paper (Marang-van de Mheen, 1998) and also to that used by
Pamuk as early as 1985 to analyze trends in England and Wales from 1921 through 1972
(Pamuk, 1985; Wagstaff, Paci & Van Doorslaer, 1991). One difference is that Pamuk used a
linear model, whereas multiplicative models are more widely used now.
The results given by the models are consistent with findings from graphical descriptive
approaches. The SES rankings were based on the mortality data rather than on an independent
opinion about ranking, which might be debatable (Wagstaff & al., 1991). We considered that
minimal: for men, the rankings were identical for all four periods; for women the RIIs would
have been slightly smaller had the groups been ranked a priori. For most of the results, the
SES ranking included several categories of inactive, classified as unemployed or retired
irrespective of their previous occupation. This methodological choice has limitations,
especially for interpreting the results in terms of social inequalities. However, we consider
that it is well adapted to describe temporal changes which can be related to employment
policies. With the models describing global changes from 1968 through 1996 it was possible
to test the hypothesis that inequalities had increased in the period, and to quantify the
magnitude of the increase.
Socioeconomic inequalities remain large in France, especially for men. In addition to
inequalities within the working population, inequalities associated with employment status
have increased. Inactivity, in France as in other countries, remains one of the major correlates
of poor health (Virtanen, Luikkonen, Vahtera, Kivimäki & Koskencuo, 2003; Sermet &
Khlat, 2004). The observed differences are both health and social inequalities, since the
income level of most of those inactive in this age group is rather low. In a context of
increasing unemployment, two mechanisms may explain the changes observed here: increased
health selection and effects of long-term unemployment such as a low income for a long
period (Mesrine, 2000; Sermet and Khlat, 2004). More precise analyses and specific models
would be needed in order to quantify the mortality risk associated with different pathways
towards unemployment and inactivity (Cambois, 2004; Ribet et al, 2003).
One might wonder whether the increase, or a part of it, is a consequence of changes in the
classifications or changes in the legislation that lead to clearer identification of the
disadvantaged or those with severe health problems. However, the only major legislative
change over this period was that the legal age at retirement became 60 years in 1982. While it
this health selection effect is not specific to the years around 1982. Age at retirement in
France varies according to occupation and depends on individual factors including number of
years worked. Most people have some latitude (within a margin of several years) about their
age at retirement, and health reasons can play a role in their decision.
It is highly plausible that health selection in employment has increased over the period,
playing a role in exclusion from the labour market and early retirement, but we found no data
here to support that hypothesis. Nonetheless, inequalities among those working (who are
supposed to be more healthy overall) did not decrease in the period. Similar questions about
Denmark in the period 1987-1994 led to the hypothesis that health-related exclusion from the
labour market had increased, thereby reducing social class differences among the employed in
Denmark (Lissau I, Rasmussen NK, Moss Hesse N & Hesse, 2001). Occupational mobility
out of the labour market seems to increase class inequalities, however results on this topic are
scarce (Cardano, Costa & Demaria, 2004).
The results highlight the importance of the methodological aspects of comparisons between
periods, especially, the classification of those not working and the ways to quantify
inequalities. In this study the magnitude of inequalities was clearly sensitive to how those not
in the labour force were classified (as specific categories, or according to their previous
category, or excluded from the comparisons). The models used here are well adapted to
describe changes over time and provide a useful addition to purely descriptive approaches.
Inequalities were more pronounced for men, consistent with findings from other studies.
Quantifying social inequalities among women appears to be complex ( Mustard & Etches,
2003; Sacker A, Firth D, Fitzpatrick R, Lynch K & Bartley, 2000). The magnitude of the
differences between the sexes may be sensitive to the choice of the inequality measure. Our
An increase of inequalities was observed both for cancer and cardiovascular diseases, and also
for external causes for men, which suggests that there is not one single explanation to the
changes over this period in terms of proximal risk factors. The magnitude of inequalities by
causes of death depends on the country; changes over time for a given country reflect changes
in the social distribution of risk factors (Mackenbach, 2003; Khang, Lynch & Kaplan, 2004;
Leclerc, 1989). For France the contribution of cancer to social inequalities was large in the
period 1970-1980 (Leclerc, 1989); it remains large, especially for men (Menvielle, Luce,
Geoffroy-Perez, Chastang & Leclerc, 2005). In other European countries widening
inequalities in the period 1983-1993 were also observed for lung cancer (Mackenbach & al,
2003). The increase of inequalities for cardiovascular mortality in France has been described
and discussed (Lang & Ducimetiere, 1995). The authors indicated that disparities between
active and inactive people had increased, which suggested the role of a health-related
selection process towards unemployment. For “other causes of deaths” social inequalities
were larger in France than in other European countries in the period 1970-1980 (Leclerc,
1989).
The magnitude of the inequalities given here for France cannot easily be compared with
results from other studies. Some of these compare non-manual to manual workers
(Mackenbach & al., 2003; Donkin & al, 2002); others assess SES according to educational
level (Khang YH, Lynch JW, Yun S, Lee, 2004) or income (Davey Smith & al., 2002), or
according to occupation at death, rather than at the census as in our study (Marang-Van de
Mheen & al., 1998). Comparisons between countries are sensitive to methodological choices,
such as how SES is defined for those not working. It is also necessary to take into account the
age differences between studies, since the magnitude of inequalities varies with age.
In many developed countries labour market flexibility has increased, which implies
2001; Virtanen, Luikkonen, Vahtera, Kivimäki & Koskencuo, 2003). One of the questions
raised here is to which extend labour market flexibility increases social inequalities in
mortality. It would be difficult to quantify this phenomena with precision. At least, one
conclusion is that traditional description of inequalities limited to employed groups can be
misleading, since the labour market status is closely associated with health (Lissau,
Rasmussen, Moss Hesse & Hesse, 2001). Specific studies are needed to improve our
understanding of the effects of changes in the labour market, such as increased
unemployment, on various segments of the population and to quantify these effects on
mortality.
Appendix - Global model and estimates of the absolute differences for mortality.
In the models presented in Table 5 and Figure 3, the reference (100 in figure 3) is the
mortality of a person with the average SES at the first period. SES was coded on a scale from
-0.5 to 0.5. This coding was preferred to a coding from 0 to 1, because interaction terms were
added to the model. Details of the estimation of absolute differences according to the model
are given below for men; The figures for “mortality” in figure 3 are not mortality rates, but
standardised mortality ratios, with the reference (100) as mentioned above.
The RII for the first period is 3.937. Hence the estimate of mortality at the bottom of the
social scale for this period is 100 x (3.937) ½ = 198.4. The estimate at the top of the social
scale for the same period is 100 x (3.937) -½ = 50.4. The difference is 148.
For the most recent period, the mortality for someone with an average SES is 100 x 0.742,
and the RII is 3.937 x 1.468. Hence the estimates for the bottom and the top of the social scale
are:
100 x 0.742 x (3.937 x 1.468) -½ = 30.86
The difference is 147.5.
The reference 100 might be replaced by the standardized mortality rate at the middle of the
social scale for the first period, which is not exactly the mortality rate of the general
population, because the relation between the relative social situation and mortality is not
linear in the model. Hence there is a slight difference between the mortality rate of the general
population, which is average mortality, and the mortality of an “average” person (according to
his or her SES).
References
Bartley M, Ferrie J. (2001). Glossary: unemployment, job insecurity, and health. J Epidemiol
Community Health, 55, 776-781.
Bostrom G, Rosen M. (2003). Measuring social inequalities in Health – politics or science?
Scand J Public Health, 31, 3, 211-215.
Brahimi M. (1980). La mortalité des étrangers en France. Population,3,603-622.
Cambois E. (2004). Occupational and educational differentials in mortality in French elderly people : magnitude and trends over recent decades. Demographic research, Special collection 2, article 11:277-304.
Cambois E. (2004). Careers and mortality in France: evidence on how far occupational mobility predicts differentiated risks. Social Science and Medicine 58,12, 2545-2558.
Cardano M, Costa G & Demaria M. (2004). Social mobility and health in the Turin longitudinal study. Social Science and Medicine, 58,8, 1563-1574.
Chenu A. Le repérage de la situation sociale.(2000). In Leclerc A, Fassin D, Grandjean H, Kaminski M & Lang T (Eds.) les inégalités sociales de santé (pp 93-108). Paris: La Découverte/INSERM.
Davey Smith G, Dorling D, Mitchell R & Shaw M. (2002). Health inequalities in Britain: continuing increases up to the end of the 20th century. J Epidemiol Community Health, 56, 434-435
Donkin A, Goldblatt P & Lynch K. (2002). Inequalities in life expectancy by social class, 1972-1999. Health statistics Quarterly 15, 5-15.
Héran F. (1998). EDP, présentation générale. Economie et Statistique 6/7, 316-317.
Khang YH, Lynch JW, Yun S & Lee SI. (2004).Trends in socioeconomic health inequalities in Korea: use of mortality and mortality measures. J Epidemiol Community Health 58, 4, 308-314.
Khang YH, Lynch JW & Kaplan GA. (2004). Health inequalities in Korea: age - and sex – specific educational differences in the 10 leading causes of deaths. International Journal of
Epidemiology 33:299-308.
Kunst AE. (1997). Cross-national comparisons of socio-economic differences in mortality. Thesis Erasmus University Rotterdam, The Netherlands, 264p.
Lang T, Fassin D, Grandjean H, Kaminski M & Leclerc A. France. (2002). In Mackenbach J & Bakker M (Eds.), Reducing inequalities in health; an European perspective (pp 214-222). London: Routledge.
Lang T, Ducimetiere P. (1995). Premature cardiovascular mortality in France: divergent evolution between social categories from 1970 to 1990. International Journal of
Leclerc A. (1989). Differential mortality by cause of death: comparison between selected European countries. In Fox J (Ed.), Health inequalities in European countries (pp 92-108). Aldelrshot: Gower.
Leclerc A, Lert F & Fabien C. (1990). Differential mortality: some comparisons between England and Wales, Finland and France, based on inequalities measures. International
Journal of Epidemiology 19,4, 1-10.
Lissau I, Rasmussen NK, Moss Hesse N & Hesse U. (2001). Social differences in illness and health-related exclusion from the labour market in Denmark from 1987 to 1994. Scandinavian
Journal of Public Health S55, 19-30.
Mesrine A. (1999). Les différences de mortalité par milieu social restent fortes. Données
sociales, 228-235.
Mesrine A. (2000). La surmortalité des chômeurs : un effet catalyseur du chômage ?
Economie et Statistique 334,4, 33-48.
Mackenbach JP & Kunst AE. (1997). Measuring the magnitude of socio-economic inequalities in health : an overview of available measures illustrated with two examples from Europe. Social Science and Medicine 44, 6, 757-771.
Mackenbach JP, Kunst AE, Cavelaars EJM, Groenhof F, Geurts JJM & the EU Working Group on Socio-economic Inequalities in Health.(1997). Socio-economic inequalities in morbidity and mortality in western Europe. The Lancet 349, 1655-1659.
Mackenbach JP, Bos V, Andersen O, Cardano M, Costa G, Harding S, Reid A, Hemström O, Valkonen T & Kunst AE. (2003). Widening socioeconomic inequalities in mortality in six Western European countries. International Journal of Epidemiology 32, 830-837.
Marang-Van de Mheen PJ, Davey Smith G, Hart CL & Gunnings-Schepers LJ. (1998). Socio-economic differentials in mortality among men within Great Britain : time trends and contributory causes. J Epidemiol Community Health 52, 4, 214-218.
Menvielle G, Luce D, Geoffroy-Perez B, Chastang J-F, Leclerc A & the EDISC group.(2005). Social inequalities and cancer mortality in France. Cancer causes and control 16, 501-513.
Monteil C & Robert-Bobée I. (2005). Les différences sociales de mortalité : en augmentation chez les hommes, stables chez les femmes. INSEE première 1025, 1-4.
Mustard CA & Etches J. (2003). Gender differences in socioeconomic inequality in mortality.
J Epidemiol Community Health 57, 974-980.
Pamuk ER. (1985). Social class and inequality in mortality from 1921 to 1972 in England and Wales. Population Studies 39, 17-31.
Regidor E. (2004). Measures of inequalities: part 2. J Epidemiol Community Health 58, 900-903.
Ribet C, Zins M, Gueguen A, Bingham A, Goldberg M, Ducimetiere P, Lang T. (2003). Occupational mobility and risk factors in working men: selection, causality, or both? J
Epidemiol Community Health 57, 901-906.
Rouault D. (1994). The EDP – A French equivalent to the Longitudinal Study. Updated news from the LS User Group 7:5-13.
Sacker A, Firth D, Fitzpatrick R, Lynch K & Bartley M. (2000). Comparing health inequality in men and women: prospective study of mortality 1986-96. BMJ 320, 1303-7.
Sermet C, Khlat M. (2004). La santé des chômeurs en France : revue de la littérature. Revue
d’Epidémiologie et Santé Publique, 52, 5, 465-474.
Singh-Manoux A, Ferrie JE, Lynch JW & Marmot M. (2005). The role of cognitive ability (intelligence) in explaining the association between socioeconomic position and health: Evidence from Whitehall II prospective cohort study. American Journal of
Virtanen P, Luikkonen V, Vahtera J, Kivimäki M & Koskencuo M. (2003). Health inequalities in the workforce: the labour market core-periphery structure. International
Journal of Epidemiology 32,6, 1015-1021.
Wagstaff A, Paci P & Van Doorslaer E. (1991). On the measurement of inequalities in health.